Prediction of hERG potassium channel PAS-domain variants trafficking via machine learning
1Congenital long QT syndrome (LQTS) is characterized by a prolonged QT-interval on an electrocardiogram (ECG). An abnormal prolongation in the QT-interval increases the risk for fatal arrhythmias despite otherwise normal metrics of cardiac function. Genetic variants in several different cardiac ion channel genes, including KCNH2, are known to cause LQTS. The population frequency of rare non-synonymous (missense) variants in LQTS-linked genes far outpaces the true incidence of the disease. Therefore, only a small percentage of missense variants identified in LQTS-linked genes are expected to associate with LQTS. Because of a lack of clear association between variants identified in LQTS-linked alleles and diseases, most variants are classified as variants of uncertain physiological significance (VUS). Here, we evaluated whether structure-based molecular dynamics (MD) simulations and machine learning (ML) can improve the identification of missense variants in LQTS-linked genes that associate with LQTS. To do this, we focused on investigating KCNH2 missense variants in the Kv11.1 channel protein shown to have wild type (WT) like or loss-of-function (LOF) phenotypes in vitro. We focused on KCNH2 missense variants that disrupt normal Kv11.1 channel protein trafficking, as it is the most common LOF phenotype for LQTS-associated variants. Specifically, we used these computational techniques to correlate structural and dynamic changes in the Kv11.1 channel protein PAS domain (PASD) with Kv11.1 channel protein trafficking phenotypes. These simulations unveiled several molecular features, including the numbers of hydrating waters and H-Bonds, as well as FoldX scores, that are predictive of trafficking. We then used statistical and ML (Decision tree (DT), Random forest (RF), and Support vector machine (SVM)) techniques to classify variants using these simulation-derived features. Together with bioinformatics data, such as sequence conservation and folding energies, we were able to predict with reasonable accuracy ({approx}75%) which KCNH2 variants do not traffic normally. We conclude, structure-based simulations of KCNH2 variants localized to the Kv11.1 channel PASD led to a significant improvement ({approx}10%) in classification accuracy and this approach should therefore be considered to complement the classification of VUS in the Kv11.1 channel PASD.